We usually have only limited knowledge about the economic consequences of land-use decisions, thus they are uncertain. We analyze the implications of this uncertainty on conservation payments (CP) to preserve wildlife-friendly shade coffee production in southwest Ecuador, when conversion to maize is the most profitable alternative. Our objective is twofold: First, we analyze the consequences of applying Stochastic Dominance (SD) to derive CP, an approach making only minimal assumptions about the preferences of farmers. Second, we investigate the effects of land-use diversification to reduce CP by allowing for shade coffee on part of a landholding, and maize production on what remains. CP derived by SD turned out to be at least twice the amount calculated by an alternative method which maximizes a concave utility function-US$ 166 to US$ 294 ha -1 year -1 instead of US$ 86 ha -1 year -1 . Given this result, we doubt that the assumptions underlying SD are reasonable for farmers, who are known to be riskaverse. Allowing for land-use diversification has a significant impact on CP. The optimal portfolio share of shade coffee is 27 % and for maize 73 % for moderately risk-averse farmers-without any CP. A larger share of shade coffee is preferable for strongly risk-averse farmers-51 and 49 % maize. The amount of CP necessary to encourage the expansion of shade coffee to 75 % is US$ 40 ha -1 year -1 (for moderately risk-averse) and US$ 19 ha -1 year -1 (for strongly risk-averse farmers). Stimulating diversification may thus help to significantly reduce CP necessary to preserve less profitable agroforestry options.